An increasing number of application use-cases such as Internet-of-Things (IoT), industrial automation, Augmented Reality (AR), smart cities, autonomous trans-portation systems, etc., are warranting the need for edge compute clouds [102–104].
Several other research areas, such as Industry 4.0 utilize edge clouds to provide
97 5.6 Conclusion
efficient solutions to several open questions. For example, researchers have pro-posed automated collaborative robots which require time-critical processing with extremely low latency (in order of milliseconds) to create a safety zone for their op-erators [92,105]. Augmented Reality glasses can assist operators in a continuously varying production environment by performing markerless object recognition and accurate tracking in a factory. However, such use-cases can only be fulfilled if the required data is cached and computed at closely located servers.
State of the art solutions in the fields of virtualization, Software Defined Network-ing (SDN) and Network Function Virtualization (NFV) represents key technologies to deploy virtualized services at the very edge of the network in a flexible way and on cheap commodity hardware [91]. In this paper, we envisioned the case in which clients have control over which services will be included in their network path using state of the art Service Function Chaining (SFC) techniques such as [35,37]. In an open market of choosing services, the client can discover specific edge servers which are hosting the required service by utilizing Domain Name System (DNS) based techniques [78,106]. The clients can significantly enhance their connectivity with the end-server by using services with very low network delay.
5.6 Conclusion
In this chapter, we proposedMute, a multi-tier edge cloud architecture which enables edge cloud providers to efficiently deploy services at the edge. Mute categorizes edge servers into groups based on their network delay from the client. Due to its unique architecture abstraction,Mute can efficiently deploy service function chains on edge servers across multiple edge platforms. Through our extensive
simulation-Chapter 5 Mute: MUlti-Tier Edge networks 98
based evaluation on RocketFuel topologies, we show thatMute achieves a significant reduction in edge network delay and completion time when compared to state of the art.
In our future work, we plan to investigate the impact of different edge resources clustering strategies on the service placement.
Chapter 6
Conclusion & Future Work
In this thesis, we explored the solutions available in the state of the art with regards to SFC systems, analyzing their characteristics and limitation. Along these lines, we introduced a high-level categorization of SFC systems based on their requirements and target use-case: single-domainandinternet-wideSFCs. The former targets SFC systems typically deployed within a single administrative domain, such as Mobile Networks and Data Centers, which are characterized by the fact that the SFC is defined and enforced by the same stakeholder. The latter instead, are characterized by the fact that multiple stakeholders are included in the SFC provisioning. Such logical separation helps in finding solutions which are optimized for the specific use-case requirements and specifications.
We observed that single-domain SFC solutions were characterized by a ”clean-state” approach, when considering the SFC system design and implementation. In fact, they do not consider prior system architecture and network equipment. In Chapter 3, we propose Catenae, a ready-to-deploy SFC solution that can be de-ployed on legacy Mobile Networks infrastructure. It provides an effective SFC sys-tem without requiring important changes in the network infrastructure. Further, Catenae scales to provide fine grained policies for millions of network flows. We proved it can handle three times the expected traffic rate at the PGW of 2019 (5 years later the publication).
In Chapter 4, we explored the state of the art for internet-wide SFC solutions,
Chapter 6 Conclusion & Future Work 100
highlighting their main characteristics and requirements. We identified an impor-tant limitation, shared among the state of the art solutions. They assume that the clients have the knowledge of the IP addresses of NFs composing the SFC prior to the connection establishment. However, no prior work is available in the state of the art regarding how it is possible to retrieve those information, and which are the challenges in such process in the current Internet scenario. Therefore, in Section4.4, we proposed to use the current Domain Name System (DNS) to retrieve the IP ad-dresses related to a SFC, highlighting its properties and inefficiencies. In Section4.4, we proposed a collaborative SFC Resolution approach. It requires minimal changes to the current DNS architecture, enabling to achieve close-to-optimum NF instance selection. The presented solution, implemented using a wide deployed system such as the DNS, represents the basis to foster the wide deployment of internet-wide SFC techniques. The presented system has been submitted as a Patent application [16]
and is currently under review.
The DNS is a well established Internet service, however, at the same time, it is always evolving and optimizing itself as its role is crucial within the Internet architecture. In fact, the DNS resolution time affects the connection establishment time for all the web traffic. Further, it has been proved that the DNS traffic accounts for more than 50% of the overall web traffic [107]. Therefore, as future work, we envision to deploy an efficient implementation of the collaborative SFC Resolution presented in Section 4.4.
New application scenarios such as Internet-of-Things (IoT), vehicular networks, etc. have proliferated to a great extent in the Internet scenario. Such applications are requiring to offload computational resources with stringent completion time re-quirements. This is the case of devices not equipped with enough computation power on-board (e.g., sensors, electrical appliance, etc.) or when the task to
per-101
form is too computational intensive (e.g., autonomous driving). In either cases, the traditional cloud model might fail to support those use cases due to possibly high network delays encountered while offloading data to the location of cloud data cen-ters. Therefore, traditional cloud resources are being decoupled in smaller resources, deployed closer to the users, Due to their proximity to the network ”edge”, these collections of resources are termed as Edge cloud.
We investigated the state of the art with regards to edge computing solutions. SFC and Edge Computing are related by the fact that a SFC is usually implemented from the client (e.g., IoT device, car, etc.), through an edge deployed service, to the cloud deployed server. However, the state of the art with regards to the service placement on the edge is very limited, as it is only consider the case in which there are homo-geneous edge resources distributed in few location in the network. In Chapter5, we proposedMute, a multi-tier edge cloud architecture in which several heterogeneous edge resources are widely distributed in the network. Mute enables edge cloud providers to efficiently deploy services on such network architecture, categorizing edge servers into groups, based on their characteristics (e.g., network delay from the client, resource availability, etc.). Due to its unique architecture abstraction, Mute achieves a significant reduction in edge network delay and completion time when compared to state-of- the-art solutions, when applied to this infrastructure. In the context of multi-tier edge networks, we plan to investigate the impact of different edge resources clustering strategies on the service placement.
Chapter 6 Conclusion & Future Work 102
6.1 Thesis impact
Scientific Publications The work on designing, implementing and evaluating Catenae has been published in the following peer-reviewed international conference proceeding.
• Roberto Bifulco, Anton Matsiuk, and Alessio Silvestro. ”Ready-to-deploy service function chaining for mobile networks.” NetSoft Conference and Work-shops (NetSoft). IEEE, 2016 [14].
An extension of Catenae that includes the hw-sw switch design and implementa-tion has been published in the following peer-reviewed internaimplementa-tional journal.
• Roberto Bifulco, Anton Matsiuk, andAlessio Silvestro. CATENAE: A scal-able Service Function Chaining system for legacy mobile networks. Interna-tional Journal of Network Management, 27(2), 2017 [15].
The preliminary work and the full system architecture on enabling Internet-wide SFC has been published in the following peer-reviewed international conference proceed-ings.
• Alessio Silvestro, Roberto Bifulco, Fabian Schneider, Xiaoming Fu, and Jussi Kangasharju. ”Is today’s DNS the right solution for middleboxes se-lection?”. Proceedings of the 4th Workshop on CrossCloud Infrastructures &
Platforms. ACM, 2017. [Poster] [78].
• Alessio Silvestro, Roberto Bifulco, Fabian Schneider, Xiaoming Fu, and Jussi Kangasharju. ”MISE: MIddleboxes SElection for multi-domain service function chains.” Proceedings of the 2nd Workshop on Cloud-Assisted
Net-103 6.1 Thesis impact
working. ACM, 2017 [106].
Thus, the work on enabling Multi-tier Edge Networks has been published in the following peer-reviewed international conference proceeding.
• Alessio Silvestro, Nitinder Mohan, Jussi Kangasharju, Fabian Schneider, and Xiaoming Fu. MUTE: MUlti-Tier Edge networks. In Proceedings of the 5th Workshop on CrossCloud Infrastructures & Platforms (CrossCloud’18).
ACM, 2018. [108]
Patent Applications Further, related to the broad topic of this dissertation, preliminary results and ideas have been published as patent applications.
• Alessio Silvestro, Dirk Kutscher, and Fabian Schneider. Method and System for Introducing In-Network Services in an End-To-End Communication Path.
Publication number: WO 2017/194168 A1. Application date: 2016-05-13.
Publication date: 2017-11-16. [109]
• Fabian Schneider, Alessio Silvestro, and Thomas Dietz. Software Defined Network and Method for Operating the same. Application date: 2016-12-22.
A method to flow installation time sensitive network control [110].
• Alessio Silvestro, Fabian Schneider, and Roberto Bifulco. Explicit Service Function Chaining (SFC) using DNS extensions. Application date: 2017-03-10. [16]
The text of [109] is publicly available. [16,110] instead, are still under review and the Patent applications text cannot be made public. However, I would like to invite Ph.D. committee members, who are interested in accessing those documents, to sign an Non-Disclosure Agreement (NDA) with NEC Laboratories Europe in order to
Chapter 6 Conclusion & Future Work 104
obtain a copy of the Patents application text.
Bibliography
[1] David Clark. The design philosophy of the darpa internet protocols. ACM SIGCOMM Computer Communication Review (CCR), 18(4):106–114, 1988.
[2] Jerome H Saltzer, David P Reed, and David D Clark. End-to-end arguments in system design. ACM Transactions on Computer Systems (TOCS), 2(4):277–
288, 1984.
[3] Cisco Visual Networking Index: Forecast and Methodology, 2016–2021.
September 15, 2017. [Online]. Available: https://www.cisco.com/c/en/us/
solutions/collateral/service-provider/visual-networking-index-vni/complete-white-paper-c11-481360.html.
[4] [Published: 07.12.2015]. [On-line accessed: 22.05.2018]. [Online]. Avail-able: https://venturebeat.com/2015/12/07/streaming-services-now- account-for-over-70-of-peak-traffic-in-north-america-netflix-dominates-with-37/.
[5] IBM Institute for Business Value, “Five telling years, four future scenarios,”
2010.
[6] Jim Liddle. Amazon found every 100ms of latency cost them 1% in sales. The GigaSpaces, 27, 2008.
[7] Ibrar Yaqoob, Ejaz Ahmed, Muhammad Habib ur Rehman, Abdelmuttlib Ibrahim Abdalla Ahmed, Mohammed Ali Al-garadi, Muhammad Imran, and Mohsen Guizani. The rise of ransomware and emerging security challenges in the Internet of Things. Computer Networks, 129:444–458, 2017.
Bibliography 106
[8] Theophilus Benson, Aditya Akella, and David A Maltz. Unraveling the Com-plexity of Network Management. InNSDI, pages 335–348, 2009.
[9] Ali Ghodsi, Scott Shenker, Teemu Koponen, Ankit Singla, Barath Raghavan, and James Wilcox. Intelligent design enables architectural evolution. In Pro-ceedings of the 10th ACM Workshop on Hot Topics in Networks (HotNETs’11), page 3. ACM, 2011.
[10] Raghavan, Barath and Casado, Mart´ın and Koponen, Teemu and Ratnasamy, Sylvia and Ghodsi, Ali and Shenker, Scott. Software-defined internet archi-tecture: decoupling architecture from infrastructure. In Proceedings of the 11th ACM Workshop on Hot Topics in Networks (HotNETs’12), pages 43–48.
ACM, 2012.
[11] Hyojoon Kim and Nick Feamster. Improving network management with soft-ware defined networking. IEEE Communications Magazine, 51(2):114–119, 2013.
[12] Jianli Pan, Subharthi Paul, and Raj Jain. A survey of the research on future internet architectures. IEEE Communications Magazine, 49(7), 2011.
[13] Justine Sherry, Shaddi Hasan, Colin Scott, Arvind Krishnamurthy, Sylvia Ratnasamy, and Vyas Sekar. Making middleboxes someone else’s problem:
network processing as a cloud service. ACM SIGCOMM Computer Commu-nication Review (CCR), 42(4):13–24, 2012.
[14] Roberto Bifulco, Anton Matsiuk, and Alessio Silvestro. Ready-to-deploy ser-vice function chaining for mobile networks. In NetSoft Conference and Work-shops (NetSoft), 2016 IEEE, pages 175–183. IEEE, 2016.
107 Bibliography
[15] Roberto Bifulco, Anton Matsiuk, and Alessio Silvestro. Catenae: A scalable service function chaining system for legacy mobile networks. International Journal of Network Management (IJNM’17), 27(2), 2017.
[16] Alessio Silvestro, Roberto Bifulco, and Fabian Schneider. Explicit Service Function Chaining (SFC) using DNS extensions. Patent. Application date:
2017-03-10, 2017.
[17] Nick McKeown. Software-defined networking. INFOCOM keynote talk, 17(2):30–32, 2009.
[18] Teemu Koponen, Martin Casado, Natasha Gude, Jeremy Stribling, Leon Poutievski, Min Zhu, Rajiv Ramanathan, Yuichiro Iwata, Hiroaki Inoue, Takayuki Hama, et al. Onix: A distributed control platform for large-scale production networks. In OSDI, volume 10, pages 1–6, 2010.
[19] Sushant Jain, Alok Kumar, Subhasree Mandal, Joon Ong, Leon Poutievski, Arjun Singh, Subbaiah Venkata, Jim Wanderer, Junlan Zhou, Min Zhu, et al.
B4: Experience with a globally-deployed software defined WAN. 43(4):3–14, 2013.
[20] Nick McKeown, Tom Anderson, Hari Balakrishnan, Guru Parulkar, Larry Pe-terson, Jennifer Rexford, Scott Shenker, and Jonathan Turner. Openflow:
enabling innovation in campus networks. ACM SIGCOMM Computer Com-munication Review (CCR), 38(2):69–74, 2008.
[21] Masayoshi Kobayashi, Srini Seetharaman, Guru Parulkar, Guido Appenzeller, Joseph Little, Johan Van Reijendam, Paul Weissmann, and Nick McKeown.
Maturing of openflow and software-defined networking through deployments.
Computer Networks, 61:151–175, 2014.
Bibliography 108
[22] Chi-Yao Hong, Srikanth Kandula, Ratul Mahajan, Ming Zhang, Vijay Gill, Mohan Nanduri, and Roger Wattenhofer. Achieving high utilization with software-driven wan. InACM SIGCOMM Computer Communication Review (CCR), volume 43, pages 15–26. ACM, 2013.
[23] Arpit Gupta, Laurent Vanbever, Muhammad Shahbaz, Sean P Donovan, Bran-don Schlinker, Nick Feamster, Jennifer Rexford, Scott Shenker, Russ Clark, and Ethan Katz-Bassett. Sdx: A software defined internet exchange. ACM SIGCOMM Computer Communication Review (CCR), 44(4):551–562, 2015.
[24] Arjun Singh, Joon Ong, Amit Agarwal, Glen Anderson, Ashby Armistead, Roy Bannon, Seb Boving, Gaurav Desai, Bob Felderman, Paulie Germano, et al. Jupiter rising: A decade of clos topologies and centralized control in google’s datacenter network. InACM SIGCOMM Computer Communication Review (CCR), volume 45, pages 183–197. ACM, 2015.
[25] Georgios P Katsikas. Realizing high performance NFV service chains. PhD thesis, KTH Royal Institute of Technology, 2016.
[26] Open networking foundation. https://www.opennetworking.org/.
[27] R Guerzoni et al. Network functions virtualisation: an introduction, benefits, enablers, challenges and call for action, introductory white paper. InSDN and OpenFlow World Congress, volume 1, pages 5–7, 2012.
[28] Rashid Mijumbi, Joan Serrat, Juan-Luis Gorricho, Niels Bouten, Filip De Turck, and Raouf Boutaba. Network function virtualization: State-of-the-art and research challenges. IEEE Communications Surveys & Tutorials, 18(1):236–262, 2016.
109 Bibliography
[29] OpenFlow-enabled SDN and Network Functions Virtualization. Open Net-working Foundation (ONF) Solution Brief. February 17, 2014.
[30] ETSI GS NFV 002 V1.2.1: Network Functions Virtualisation (NFV); Ar-chitectural Framework. ETSI Ind. Spec. Group (ISG) Netw. Functions Vir-tualisation (NFV), Sophia-Antipolis Cedex, France, Dec. 2014. [Online].
Available: http://www.etsi.org/deliver/etsi_gs/NFV/001_099/002/01.
02.01_60/gs_NFV002v010201p.pdf.
[31] Paul Quinn and Tom Nadeau. Problem Statement for Service Function Chain-ing. RFC 7498, April 2015.
[32] Joel Halpern and Carlos Pignataro. Service Function Chaining (SFC) Archi-tecture. RFC 7665, October 2015.
[33] W. Haeffner et al. SFC Use Cases in Mobile Networks IETF Draft.
[34] S. Kumar et al. SFC Use Cases In Data Centers. IETF Draft.
[35] David Naylor, Kyle Schomp, Matteo Varvello, Ilias Leontiadis, Jeremy Black-burn, Diego R L´opez, Konstantina Papagiannaki, Pablo Rodriguez Rodriguez, and Peter Steenkiste. Multi-context TLS (mcTLS): Enabling secure in-network functionality in TLS. InACM SIGCOMM Computer Communication Review (CCR), volume 45, pages 199–212. ACM, 2015.
[36] C. Filsfils, S. Previdi, L. Ginsberg, B. Decraene, S. Litkowski, and R. Shakir.
Segment Routing Architecture. IETF Draft.
[37] Pamela Zave, Ronaldo A Ferreira, Xuan Kelvin Zou, Masaharu Morimoto, and Jennifer Rexford. Dynamic service chaining with dysco. In Proceedings of the Conference of the ACM Special Interest Group on Data Communication,
Bibliography 110
pages 57–70. ACM, 2017.
[38] Jon Matias, Jokin Garay, Nerea Toledo, Juanjo Unzilla, and Eduardo Jacob.
Toward an sdn-enabled nfv architecture. IEEE Communications Magazine, 53(4):187–193, 2015.
[39] Michio Honda, Yoshifumi Nishida, Costin Raiciu, Adam Greenhalgh, Mark Handley, and Hideyuki Tokuda. Is It Still Possible to Extend TCP? In ACM IMC ’11, 2011.
[40] Zhaoguang Wang, Zhiyun Qian, Qiang Xu, Zhuoqing Mao, and Ming Zhang.
An untold story of middleboxes in cellular networks. InACM SIGCOMM ’11.
[41] ETSI. Network Functions Virtualisation - White Paper. [Online]. Available:
https://portal.etsi.org/Portals/0/TBpages/NFV/Docs/NFV_White_
Paper3.pdf.
[42] IETF. Service Function Chaining working group. SFC.
[43] Sevil Mehraghdam, Matthias Keller, and Holger Karl. Specifying and placing chains of virtual network functions. In Cloud Networking (CloudNet), 2014 IEEE 3rd International Conference on, pages 7–13. IEEE, 2014.
[44] Zafar Ayyub Qazi, Cheng-Chun Tu, Luis Chiang, Rui Miao, Vyas Sekar, and Minlan Yu. SIMPLE-fying middlebox policy enforcement using SDN. ACM SIGCOMM computer communication review. 43(4):27–38, 2013.
[45] Seyed Kaveh Fayazbakhsh, Luis Chiang, Vyas Sekar, Minlan Yu, and Jeffrey C Mogul. Enforcing Network-Wide Policies in the Presence of Dynamic Middle-box Actions using FlowTags. In NSDI, volume 14, pages 533–546, 2014.
111 Bibliography
[46] J. Blendin, J. Ruckert, N. Leymann, G. Schyguda, and D. Hausheer. Position paper: Software-defined network service chaining. InIEEE EWSDN ’14.
[47] Ying Zhang, Neda Beheshti, Ludovic Beliveau, Geoffrey Lefebvre, Ravi Manghirmalani, Ramesh Mishra, Ritun Patneyt, Meral Shirazipour, Ramesh Subrahmaniam, Catherine Truchan, et al. Steering: A software-defined net-working for inline service chaining. In Network Protocols (ICNP), 2013 21st IEEE International Conference on, pages 1–10. IEEE, 2013.
[48] P.Quinn, U.Elzur, and C.Pignataro. Network Service Header. IETF RFC8300.
January 2018.
[49] Mohamed Boucadair, Christian Jacquenet, Yuanlong Jiang, Ron Parker, and Kengo. Requirements for service function chaining (sfc). Internet-Draft draft-boucadair-sfc-requirements-06, IETF Secretariat, February 2015.
[50] Dan Levin, Marco Canini, Stefan Schmid, and Anja Feldmann. Incremen-tal sdn deployment in enterprise networks. In ACM SIGCOMM Computer Communication Review (CCR), volume 43, pages 473–474. ACM, 2013.
[51] Joao Martins, Mohamed Ahmed, Costin Raiciu, Vladimir Olteanu, Michio Honda, Roberto Bifulco, and Felipe Huici. ClickOS and the Art of Network Function Virtualization. In USENIX NSDI ’14, 2014.
[52] Stefano Vissicchio, Laurent Vanbever, and Jennifer Rexford. Sweet little lies:
Fake topologies for flexible routing. InProceedings of the 13th ACM Workshop on Hot Topics in Networks, page 3. ACM, 2014.
[53] Xing Xu, Yurong Jiang, Tobias Flach, Ethan Katz-Bassett, David Choffnes, and Ramesh Govindan. Investigating Transparent Web Proxies in Cellular
Bibliography 112
Networks. In PAM. Springer, 2015.
[54] Junxian Huang, Feng Qian, Yihua Guo, Yuanyuan Zhou, Qiang Xu, Z. Morley Mao, Subhabrata Sen, and Oliver Spatscheck. An in-depth study of LTE:
Effect of network protocol and application behavior on performance. InACM SIGCOMM ’13.
[55] Xin Jin, Li Erran Li, Laurent Vanbever, and Jennifer Rexford. Softcell: Scal-able and flexible cellular core network architecture. InProceedings of the ninth ACM conference on Emerging networking experiments and technologies, pages 163–174. ACM, 2013.
[56] Philipp Richter, Mark Allman, Randy Bush, and Vern Paxson. A primer on ipv4 scarcity. ACM SIGCOMM Computer Communication Review (CCR), 45(2):21–31, 2015.
[57] Ben Pfaff, Justin Pettit, Teemu Koponen, Ethan Jackson, Andy Zhou, Jarno Rajahalme, Jesse Gross, Alex Wang, Joe Stringer, Pravin Shelar, Keith Ami-don, and Martin Casado. The design and implementation of open vswitch. In USENIX NSDI ’15.
[58] Nanxi Kang, Zhenming Liu, Jennifer Rexford, and David Walker. Optimiz-ing the ”one big switch” abstraction in software-defined networks. In ACM CoNEXT’13.
[59] Maurizio Dusi, Roberto Bifulco, Francesco Gringoli, and Fabian Schneider.
Reactive logic in software-defined networking: Measuring flow-table require-ments. In Proceedings of the 5th International Workshop on TRaffic Analysis and Characterization (TRAC), 2014.
113 Bibliography
[60] Aggelos Lazaris, Daniel Tahara, Xin Huang, Li Erran Li, Andreas Voellmy, Y. Richard Yang, and Minlan Yu. Tango: Simplifying SDN Programming with Automatic Switch Behavior Inference, Abstraction, and Optimization.
In ACM CoNEXT’14.
[61] Roberto Bifulco and Anton Matsiuk. Towards scalable sdn switches: Enabling faster flow table entries installation. In ACM SIGCOMM Computer Commu-nication Review (CCR), volume 45, pages 343–344. ACM, 2015.
[62] Naga Katta, Omid Alipourfard, Jennifer Rexford, and David Walker. Infinite cacheflow in software-defined networks. In Proceedings of the third workshop on Hot topics in Software Defined Networking (HotSDN’14), pages 175–180.
ACM, 2014.
[63] Open Networking Foundation. Openflow specification 1.4.0.
[64] Nadi Sarrar, Steve Uhlig, Anja Feldmann, Rob Sherwood, and Xin Huang.
[64] Nadi Sarrar, Steve Uhlig, Anja Feldmann, Rob Sherwood, and Xin Huang.